Let the research keep the limits that make it useful
I make audience-research data explain who answered, how they answered and which population the evidence can reasonably describe before a persuasive chart enters a client decision.
01 / Chief Data Officer
Data organisation
40 colleagues
Five leads across research data and product definitions
Annual function expense
₹7.8 crore
Includes ₹2.2 crore improvement work
Survey and data career
30 years
Survey support through audience-research data leadership
＋
Designed for consequence.
The most polished chart can still answer the wrong question if an interview becomes a person, a change of method becomes a trend or a weighted sample becomes a population it never observed.
Let the research keep the limits that make it useful
I once approved a report describing 2,400 pooled interview records as 2,400 different people. The records came from repeated research waves, and some respondents had participated more than once. There were 1,650 distinct respondents in the selected pool. The interviews were real within the fictional study; the claim I approved had changed the unit being counted. We spent ₹39 lakh over eight weeks correcting the account and the review that allowed it. That correction did not establish population reach or make the sample representative by itself.
I am Mary Varnelvi, chief data officer of Lersavni Audience Research in Shillong. Our 321 colleagues produce commissioned audience studies and syndicated reports, with ₹98 crore FY 2025–26 revenue. I lead forty data colleagues through five leads. Nitin is our chief executive and Zoya leads methodology. Our work becomes valuable when a client can understand the question the evidence supports, including the conditions that make a comparison uncertain or a wider claim inappropriate.
The later decisions concern comparability and coverage. A move from telephone interviews to web collection could not simply extend an eight-wave trend line. A paired comparison revealed response differences without proving that all of them were caused by collection mode. An online panel recruited in eighteen towns could not acquire thirty-two-district coverage through demographic weighting. We revised the products and propositions while preserving the narrower evidence they could still offer. My role is an executive data appointment, with neither a statutory seat nor a personal ownership interest. A possible first outside seat would require further governance learning, specific assessment and a realistic twenty-two-day annual commitment. I would bring experience in making an attractive data account explain what it counts, how it was collected and whose experience remains outside it.
Interview records separated from distinct respondents
The selected pooled file contained 2,400 interviews from 1,650 distinct respondents; the correction did not establish population reach.
A method change kept visible in the series
After eight telephone waves, the web wave began a separately labelled series while a 120-person paired comparison informed further review.
Coverage was not created by a weight
An eighteen-town online panel was no longer presented as evidence of thirty-two-district coverage; selection limitations remained explicit.
Improvement spending inside the data remit
Five leads coordinate forty colleagues within ₹7.8 crore annual function expense, including ₹2.2 crore improvement work.
Current focus
Audience evidence with its respondent, method and coverage intact
Conversations I welcome
Research-data leadership and preparation for a first board contribution
The leadership record
Trusted with consequential decisions.
In the boardroom
Lersavni Audience ResearchExecutive data presenter; no statutory appointment · April 2022–present
Formation
Bachelor of Science in StatisticsVelrashi Statistical College · 1996
Mary made the research proposition explain the evidence it actually held. She did not let a broader geographic label stand in for field coverage the panel had never collected.
Nitin PervalkarChief Executive, Lersavni Audience Research; fictional colleagueThe full record ↗
Contact
Research-data leadership and preparation for a first board contribution
